author's personal copy

M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

ScienceDirect

www.e l sev i e r . com/ loca te /matcha r

Powder metallurgy processing and deformation

characteristics of bulk multimodal nickel

L. Farbanieca,1, G. Dirrasa,⁎, A. Krawczynskab, F. Mompioub, H. Couquec, F. Naimid,F. Bernardd, D. Tingauda

aUniversité Paris 13, Sorbonne Paris Cité, LSPM, CNRS, 99 Avenue J.B. Clément, 93430 Villetaneuse, FrancebUniversité Paul Sabatier, CEMES, CNRS, 29 rue Jeanne Marvig 31055 Toulouse, FrancecNexter Munitions, 7 route de Guerry, 18200 Bourges, FrancedInstitut Carnot de Bourgogne, UMR 6303 CNRS, Université de Bourgogne, BP 47870, 21078 Dijon, France

A R T I C L E D A T A

⁎ Corresponding author. Tel.: +33 1 49 40 34 8E-mail addresses: lfarban1@jhu.edu (L. Fa

1 Present address: Hopkins Extreme Materi

http://dx.doi.org/10.1016/j.matchar.2014.05.001044-5803/© 2014 Elsevier Inc. All rights rese

A B S T R A C T

Article history:Received 11 February 2014Received in revised form 3 May 2014Accepted 8 May 2014Available online 22 May 2014

Spark plasma sintering was used to process bulk nickel samples from a blend of threepowder types. The resulting multimodal microstructure was made of coarse (average size~135 μm) spherical microcrystalline entities (the core) surrounded by a fine-grained matrix(average grain size ~1.5 μm) or a thick rim (the shell) distinguishable from the matrix.Tensile tests revealed yield strength of ~470 MPa that was accompanied by limited ductility(~2.8% plastic strain). Microstructure observation after testing showed debonding atinterfaces between the matrix and the coarse entities, but in many instances, shallowdimples within the rim were observed indicating local ductile events in the shell.Dislocation emission and annihilation at grain boundaries and twinning at crack tip werethe main deformation mechanisms taking place within the fine-grained matrix as revealedby in-situ transmission electron microscopy. Estimation of the stress from loop's curvatureand dislocation pile-up indicates that dislocation emission from grain boundaries and grainboundary overcoming largely contributes to the flow stress.

© 2014 Elsevier Inc. All rights reserved.

Keywords:Multimodal nickelSpark plasma sinteringEBSDIn-situ TEMPlasticity

1. Introduction

The mechanical behaviour and deformation mechanisms ofnanocrystalline (NC) materials have been studied extensivelyfor a few decades from both experimental and theoreticalpoints of view. See for example refs [1–8]. In general, whateverthe mode of processing, all of the studies, except for somespecial cases such as nano-twinned nickel [9–11], show thatthe performance obtained in terms of yield stress or strengthis at the expense of tensile ductility, which consequentlylimits their formability. Therefore, different strategies have

8; fax: +33 1 49 40 39 38.rbaniec), dirras@univ-parals Institute, Johns Hopki

8rved.

been proposed to improve the ductility of NC materials [12].Among them, the implementation of bi- or multi-modalmicrostructures that incorporate more or less importantvolume fraction of coarse-grained (CG) particles into NC orultrafine-grained (UFG) components has been suggested. Tothis end, powder metallurgy (PM) route is often used becauseof its versatility and successfully completed bimodal micro-structures [13–17]. However, in such an intricate microstruc-ture, it is often difficult to know exactly the contribution ofeach component on the overall mechanical behaviour. Also,the underlying deformation mechanisms of these components

is13.fr (G. Dirras).ns University, 3400 North Charles Street, Baltimore, MD 21218, USA.

author's personal copy

127M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

cannot be investigated with the same tools. Intuitively, it canbe claimed that the smaller grains will contribute to thestrength increase, while the coarse grains to the ductilityimprovement. However, in a recent review, Estrin andVinogradov [18] observed that bimodal microstructures proc-essed by Equal Channel Angular Pressing do not lead to theductility gain. The latter properties also depend on theprocessing route. For instance, bimodal microstructures proc-essed from blends of nanometre-sized or UFG powder particleswith CG ones may suffer grain boundary cracking [19] becauseof a residual porosity or contamination of powders, whichtherefore reduces the tensile ductility. Deformation incompat-ibility of different phases may also contribute to this effect.Nevertheless, PM techniques remain versatile routes forinnovative microstructure design, and optimisation of theseprocesses is required to better understand the deformationmechanisms that occur at different length scales or within thedifferent constituents of themicrostructure, and their potentialinteractions.

Nickel is a relatively abundant element found on theearth's crust that is used as the base in many alloycompositions intended for aeronautics, medicine or defenseapplications. In a pure form, e.g. electrodeposited, it has foundan application in the field of micro-electro-mechanical-systems (MEMS) [20–22]. Other applications are liner forexplosively formed projectile [23] and shielding for thecontainment of failed blades of auxiliary power units. Thesetwo applications have been motivated by its attractivemechanical properties, such as good ductility, high toughnessand relatively low work-hardening rate. Cold worked tomoderately high strength levels still maintain ductility.Therefore, the purpose of this study was to process andinvestigate mechanical properties of nickel samples withtailored microstructure that have enhanced strength and thedesired ductility. To this end, electron backscatter diffraction(EBSD) and in-situ transmission electron microscopy (in-situTEM) investigations were performed to analyse the deforma-tion features occurring in multimodal nickel (M-Ni) processedby spark plasma sintering (SPS) of nanometre- andmicrometre-sized powder particle blends. In the processedmicrostructure, special attention has been paid to thecontribution of the fine-grained (FG) matrix component tothe plastic deformation.

Fig. 1 – SEM images of the starting powders: (a) ANi50;(b) ANi100 and (c) AANi powders.

2. Experimental Procedures

2.1. Starting Powders

Two heterogeneous nanopowders (purity > 99.7%) havingrespectively a mean particle size of ~50 nm (ANi50) and~100 nm (ANi100) were mixed with a coarse-grained (puri-ty > 99.8%) powder aggregate (AANi) having particle size inthe range of 100–150 μm. The ANi-type nanopowders weresupplied by Argonide Corporation (Sanford, FL, USA) andfabricated by the so-called ELEX™ process [24], while AlfaAesar (Ward Hill, MA, USA) supplied the AANi powder.Scanning electron microscopy (SEM) micrographs of thepowders are presented in Fig. 1a–c. The blend was made of200 g, 300 g, and 130 g of ANi50, ANi100 and AANi powders,

respectively. The powder mixing was performed underprotective atmosphere in a glove box and was homogenizedusing a Turbula™ mixer with helicoidally movement for 15 h.Following the mixing operation, the powder blend was coldpressed under a pressure of about 25 MPa in argon atmo-sphere to minimize powder oxidation before sintering.

author's personal copy

128 M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

2.2. SPS Procedure

The sintering process was carried out in a SPS apparatus(HPD125 type) from FCT™ System GmbH due to the simulta-neous application of direct current pulses of very highintensity (several thousands of amperes) and of a uniaxialpressure to the encapsulating system (Fig. A1 of Supplemen-tary material 1). The powders were sintered for 30 min using agraphite die having an outer diameter of 90 mm and height of50 mm. Actually, the powders were first cold compacted inthe die lined with graphite foil at the sintering pressure for5 min. Next, the die containing the cold-compacted samplewas positioned between graphite punches with graphite foilbetween the compact powders and the punches. The die wasthen enveloped with a carbon felt to reduce the radiation heatlosses from the outer surface. The SPS cycles were performedat a heating rate of 50 °C min−1 with a sintering temperatureof 750 °C and an axial stress of 70 MPa, see Table 1. Thetemperature was monitored by K-type thermocouples locatedin a hole, which is about 3 mm away from the sample, RTC1,and at half height of the die in 80 mm deep hole, RTC4 andRTC6, and by a pyrometer positioned about 8 mm away fromthe sample (Fig. A2 of Supplementary material 1). Typical SPSoutput and the description of the cycles are provided in theSupplementary material (Fig. B1 of Supplementary material1). Before the characterisation, the samples were polishedusing SiC 180 grit paper, next with diamond paste up to 1 μm,and finally cleaned in an ultrasonic ethanol bath to removesurface contamination from graphite foil. The relative densitywas obtained by the ratio of the bulk density of the sinteredsamples, determined using Archimedes method in distilledwater, to the theoretical density. Each value is an average ofthree measurements (Table 1). Electro-thermal and thermo-mechanical simulations, performed with current and bound-ary temperatures at the spacers as input [25,26], revealed goodtemperature homogeneity of the SPS sample with the currenttooling. Precisely, the temperature difference between themiddle and the half-radius, and the middle and the exteriorwas found to be −1% and −2%, respectively. As a result of theprocess, discs of 30 mm in diameter were obtained, allowingthe machining of tensile specimens (Fig. B2 of Supplementarymaterial 1).

2.3. Tensile Tests

The tensile tests were performed at a constant displacementrate of 0.02 mm s−1, providing a strain rate of 0.002 s−1 using aTestwell machine equipped with a load cell of capacity 50 kN.The strain was computed from the gauge length elongation.The latter was supplementary tracked by taking a photographevery 5 s of the marked initial gauge of 10 mm in length.

Table 1 – Specifications of the SPS cycles and measuredrelative densities for the two processed discs.

Sample Sinteringtemperature

(°C)

Axialstress(MPa)

Holdingtime (min)

Density(%)

MNi-1 750 70 15 97.8MNi-2 750 70 15 98.4

2.4. Microstructure Investigations

2.4.1. EBSD InvestigationsFollowing SPS process and tensile tests, the microstructureswere examined using a Zeiss Supra 40VP Scanning ElectronMicroscope equipped with fully automated electron backscat-ter diffraction (EBSD) analysis system. The EBSD crystalorientation results were analysed with OIM™ Data Analysissoftware (version 4) from TexSem™ Laboratories. The orien-tation maps were obtained using a step size of 0.1 μm. For theEBSD examination, the specimens were ground using SiC gritpapers up to SiC 4000 grade and finally polished in a colloidalsilica solution (OP-S Struers) for 30 min.

2.4.2. In-Situ TEM CharacterisationsIn-situ TEM experiments were focussed on the contribution ofthe plastic deformation characteristics of the FG componentof the processed microstructure. To this end, the specimenswere prepared from the bulk material by cutting 3 mm ×1 mm × 0.5 mm rectangles, which were then mechanicallyground. The obtained rectangles were either thinned down byelectropolishing or by a combination of ion milling by the useof Precision Ion Polishing System (PIPS™, Gatan, Inc., Model691) and a “flash polishing” using an A2 electropolishingsolution from Struers at a voltage of 7 V for 5 s at roomtemperature. The latter preparation route was applied tocircumvent the problem of the faster etching rate of the FGcomponent compared to the CG one during electropolishing.Therefore, a fast electropolishing step, referred to as “flashpolishing”, was used to remove the damaged surface layercaused by ion milling [27]. The samples were then glued to acopper grid with cyanoacrylate that was eventually placedinto a Gatan straining holder. The in-situ TEM experimentswere carried out in a JEOL 2010HC microscope operated at200 kV. Observations of the deformation mechanisms wererecorded by a MEGAVIEW III camera at 25 fps in areas wherethe hole rim is parallel to the tensile axis, i.e. where the localstress is maximum [28].

3. Results and Discussion

3.1. Characteristics of As-Processed Microstructure

Fig. 2 shows typical microstructures of the as-processed M-Nisamples. It comprises CG multi-crystalline entities (the core)that are embedded in a FG matrix. In comparison withprevious work where only ANi-type powders were used[19,29], the larger and highly subdivided CG entities (labelledCG1 in Fig. 2a–b) came from AANi powder, while the smallerand less subdivided ones (labelled CG2 in Fig. 2a–b) resultedfrom the two heterogeneous ANi-type powders.

Very often a thick shell (S) with variable thicknesssurrounding the spherical-like CG entities exists in themicrostructure as it can be seen in Fig. 2a (see the shadedregion) and Fig. 2d. Based on the SEM analysis, the size of CGcore is about 135 ± 50 μm. These CG entities are subdividedinto FG multi-crystals (average grain size of about 11.4 μm),whose grain size distribution is shown in Fig. 3a. Thegrain size distribution of the FG matrix follows a lognormal

author's personal copy

Fig. 2 – SEMmicrographs (a, b) and EBSDmaps (c, d) showing the different characteristics of the microstructure of as-processedM-Ni samples.

129M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

behaviour and is shown in Fig. 3b. The computed averagegrain size is about 1.4 μm, and is about one order ofmagnitude lower than that of the CG entities. As for theshell that occasionally surrounds the CG entities, the corre-sponding grain size distribution is presented in Fig. 3c. Anaverage grain size of 5.3 μm was computed. Actually, from alocal point of view, the grain size within the shell is variable inthat it can be lower (as can be seen from Fig. 2a and d) or larger(as can be seen in Figs. 2b and 5a) than that of the FGmatrix orthe core (C) (Fig. 2b). These differences are not fully under-stood, and it is not clear at the moment why the shell formsand why it forms only around CG1-type powder particles.Most probably the difference in purity of the powders thatwere blended together plays a role during the SPS, as thecurrent flowmust be different because of a possible differencein the powder resistivity. It is interesting to note that whensintered separately using the same SPS processing parame-ters, none of the individual powders showed such features[29]. In addition, processing by hot isostatic pressing of a blendof ANi100 and a commercial purity CG Ni powder yieldedbimodal microstructures without a shell surrounding the CG

multi-crystalline entities [13]. Therefore it can be concludedthat the observed features are due to the SPS specificity asdescribed elsewhere [30]. Additional investigations are need-ed to clarify this point.

3.2. Mechanical Properties and EBSD Analysis of the DeformedSamples

Fig. 4 shows the true stress–true plastic strain curves aftertensile test at room temperature at a strain rate of 0.002 s−1 fortwo samples with slightly different densities MNi-1 andMNi-2(see Table 1). For comparison, the stress–strain curve of a CGnickel sample (Ni-CG) is also shown. After the onset ofyielding, two regimes of hardening are observed. The firstone occurs up to about 0.7% plastic deformation where theflow stress sharply increases from 470 MPa at yielding to about550 MPa. The second and last regime occurs afterwards andcorresponds to a limited flow stress increase of less than 20 MPawithin a plastic deformation range of 0.7–2.8% (failure point).

Compared to the CG counterpart, there is an obviousincrease of the yield stress as a consequence of a high fraction

author's personal copy

Fig. 3 – Grain size distributions of the different componentsof the microstructure extracted from EBSD experiments:(a) multi-crystallites inside CG entities; (b) FG matrix and(c) the shell surrounding some CG entities.

Fig. 4 – True stress–true plastic strain plots of roomtemperature tensile tests at 0.002 s−1. Two samples from twoconsolidated discs were tested in the same conditions. Themechanical behaviour of conventional CG-Ni (average grainsize of about 25 μm) is shown for comparison.

130 M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

of the FG component as reported elsewhere [15,17,31,32]. Inaddition, contrariwise to a previous study on homogeneousUFG nickel processed via SPS and deformed in compression at

room temperature [19], the softening behaviour that occursjust after the onset of yielding is not observed. Most probably,inter-granular micro-crack growth in the FG component andtheir subsequent propagation are delayed (or blunted) by theCG entities [17,31–33]. Further, crack network may build-up,which leads to stress saturation, and consequently to finalfailure as it is observed. This scenario is in good agreementwith recent numerical studies performed on nanostructuredmetals with bimodal grain size distribution [34]. The contri-butions of micro-cracks in the plastic deformation wereaccounted for in the mechanism-based plastic model used todescribe the strength and ductility of the bimodal metals.Indeed, it was shown that the cavitation micro-crackingcontrols the fracture mechanism in bimodal metals andalloys during tensile deformation, as usually observed in NCand FG materials. Recently, a novel microstructural designwas proposed, the so-called “Harmonic Structure Design”,which is a 3D network of FG shell with CG core, that results ina combination of high strength and high ductility at the sametime [32,35,36]. Strain hardening rate analysis suggested thatthe early stage of deformation was dominated by the plasticdeformation of the stronger FG shell region [36]. Therefore, atthe early stages of deformation, it appears that the harmonicstructure deforms in the same manner as the homogeneousFG microstructure, followed by typical deformation mode forCG materials in the later stages with a large uniformdeformation. In this process, the CG component also un-dergoes plastic deformation to adjust the shape changeof the FG shell in order to maintain the integrity of thematerial. However, in the present case, where a strong non-homogeneous grain size distribution exists, deformationincompatibilities may result in cracking and cavitation in theFG component, as observed by SEM and EBSD studies.

Fig. 5 shows SEM analysis of the fractured surface of theM-Ni sample after tensile tests. In Fig. 5a, large “dimples”homogeneously distributed in the sample cross-section canbe seen. Enlarged view presented in Fig. 5b actually showsthat they correspond to deformation features occurring due to

author's personal copy

Fig. 5 – SEM images of fracture surface: (a) overall view; (b, c)enlarged view of a small area of (a).

131M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

the presence of the CG entities, which may be responsible forthe overall tensile ductility. Still, the expected ductility waslimited by a brittle-like failure and particle separation(debonding) from the matrix. A close view presented inFig. 5c shows small dimples (marked SD) in the FG matrixadjacent to and beneath the CG entities (once they have beenejected). In addition, large and shallow dimples (LD) occur in

what appear to be a periphery region or a shell around a CGentity accompanied by a shear-like fracture surface. It istherefore concluded that the shell plays an important role onthe overall plastic deformation as it may induce furthertensile ductility in the MNi samples compared to poor orzero tensile ductility observed in the bimodal case processedby our group so far [19,29]. It is possible that the debonding, asillustrated in Fig. 5c, comes from incomplete sintering. But,given the relatively higher mass density that was reachedhere, it is also possible that that grain size gradients influencethe local behaviour, i.e. the larger the gradient the higher theprobability of debonding or cracking due to deformationincompatibilities and stress concentrations as previouslyreported in [13,19]. Therefore, given the average grain size ofthe different components, the probability of debondingshould be higher between the FG matrix and the CG entitiesthan between the FG matrix and the shell. The presentedresults are in accordance with enhanced tensile ductilityfound in the case of the so-called “harmonic” structures madeof CG core surrounded by UFG shell [37,38]. It is interesting tonote that the ratio of the average grain size between the UFGshell and the CG core was the same as in the present study.

Further EBSD investigations confirm the above analysis.Fig. 6 shows EBSD maps of the specimen after failure (2.8%macroscopic plastic deformation), near the surface. In linewith presented SEM observations, it was found that both theFG matrix and the periphery of the CG entities experienced alarger amount of plastic deformation than the centre (Fig. 6a–b). This is illustrated by the presence of low angle grainboundaries (LAGBs) in red colour (Fig. 6b). The amount ofLAGBs is clearly higher within the periphery (the shell) and atthe vicinity of cracking areas. The circular-like crack isprobably due to a complex stress state that sets in the samplenear the fractured surface. Interestingly it was observed thatin the absence of a shell, the debonding phenomenonoccurred all around the CG entity as shown in the inversepole figure image (Fig. 6c).

3.3. In-Situ TEM Investigations

In-situ observations were carried out in areas of stressconcentrations, i.e. close to crack tips. Because of residualporosities, cracks developed in regions of less resistance,more likely between voids. Fig. 7a shows the crack growthobserved during straining experiments along a tensile axisoriented vertically on the images. As expected, the crackopened perpendicularly to the tensile direction and the crackpath indicates that the fracture occurred preferentially alonggrain boundaries (GBs) of FG grains. However, observations ofhighly elongated grains, indicated by arrows in Fig. 7b,testified of significant plastic deformation. Dislocation activ-ity has also been monitored inside the grain as shown below.NC grains, probably coming from the initial powder, areusually observed in the surroundings of FG grains and theymay form small porosities, due to incomplete sintering asshown in Fig. 7c and d. This explains the reason why crackspreferentially propagated along these NC grains from poros-ities to porosities. These observations are also in agreementwith reports of crack propagation in electrodeposited nickelalong voids located at grain boundaries or triple junctions [39].

author's personal copy

Fig. 6 – EBSD investigations near the fracture surface:(a–b) inverse pole image of a grain surrounded by a shell; (c) agrain without a shell.

132 M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

Fig. 7d shows in more details an area containing NC grainsclose to GBs. It is worth noting that close to these NC grains,dislocation pinning (indicated by an arrow) can be observedevidencing the presence of very small obstacles, probablyrelated to atomic impurities.

Fig. 8 and Video 1 (Supplementary material 2) illustrateplastic deformation in the grain, shown in Fig. 7c close to acrack tip (C). Under a tensile strain along the direction (T),dislocation emission arising from a source (S) has beenobserved. Although it is difficult to locate the source becauseof strong deformation close to GB; it appears to be close to atriple junction. At t = 0 s, a dislocation loop (d) is emitted inthe grain interior and eventually expands rapidly (Fig. 8a–b).Fig. 8e represents schematically the process. One part of thedislocation intersects the sample surface and leaves a trace(tr) at one sample surface that can be identified as the (111)plane, as shown on the stereographic projection in Fig. 8c. Theother part seems to be trapped close to the GB2. The visibilityof the dislocation with the [200] diffraction vector tends toindicate that the Burgers vector is either 1

2 110� �

or 12 101� �

.However, this latter yields an almost null Schmid factor.Because the glide plane normal is only 16° from the tensiledirection, the dislocation is then expected to glide with a quitelow Schmid factor M = 0.23 for a Burgers vector 1

2 110� �

. Thiscan, however be explained since local high stress concentra-tion at crack tip can arise.

Becausemechanical tests tend to attribute the first stage ofplastic deformation to the FG component, presumably inconjunction with crack development, the emission of dislo-cations from GB close to cracks followed by their piling-up isthought to be representative of the deformation mechanismin the early stage of plastic deformation. This conclusionneeds however to be moderated by the fact that observationswere carried in thin foils, which may introduce biases whencompared to bulk materials. Usual surface effects on thedislocations such as image forces or surface pinning [40–42]can be neglected in the present case because the foil thicknessis comparable to the distance between obstacles (i.e. the GBs).Crack propagation is however probably enhanced in thin foilsbecause of the easy strain relaxation out of the thin foil plane.Long-range internal stresses that may exist in the bulkmaterial are also lowered in thin foil conditions. Concerningmore specifically GBs, diffusion processes such as dislocationclimb may also be facilitated in a thin foil [43], which mayaffect the stress required to operate source and dislocationincorporation into GB. Dislocation nucleation in GB duringin-situ straining experiments in UFG aluminium has beenrecently attributed to stress concentration at GB surfacegrooves [44].

The strength needed to activate the source can beestimated locally by measuring the dislocation curvaturewhen the dislocation embryo just begins to expand. Theshape of the dislocation in the (111) plane can be obtained bydistorting Fig. 8a to correct the inclination of the glide planewith respect to the electron beam. Using anisotropic elasticcalculations, the curvature of the dislocation can then befitted by the theoretical equilibrium shape of a dislocationloop under a stress τ (through the DISDI software) [45].

The best fit (Fig. 8d) leads to τ = 110 ± 10 MPa, corre-sponding to an applied stress of σ = τ / M = 475 ± 45 MPa.This value, which is of the same order of magnitude as themeasured macroscopic yield stress at 0.002 s−1, i.e. σYS =470 MPa, tends to indicate that the stress required to activatea dislocation source largely contributes to the overall yieldstress.

author's personal copy

Fig. 7 – In-situ TEM observations in the FGmatrix showing the influence of NC grains in deformation and damagemechanisms.See text for details.

133M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

Fig. 9a and Video 2 (Supplementary material 3) show thatmechanical twinning is also possible in FG grains. A twinlamella (L) in a (111) plane and originating from a crack (C)(Fig. 9a) is clearly visible in grain G1. It appears to be stopped atthe GB between G1 and G2. Diffraction analysis shows that theGB is also a (111) twin, presumably an annealing twin. Uponstraining, the twinning dislocations have been observed tomove toward the GB (see the movie in the Supplementarymaterials). Although twinning is supposed to be disfavouredcompared to perfect dislocation emission because of theformation of high stacking fault energy, it can occur in highstress concentration area where partial nucleation is easier[46] or in NC material due to the high line-tension energy ofperfect dislocations [47]. Twinning has also been observed inNC nickel [48].

A closer inspection indicates that the lamella is in factcomposed of two 20 nm thick twins (Tw), which are separatedby a distance of about 30 nm (Fig. 9b). A zoom in the area ofintersection between the twins shows that extrinsic disloca-tions have been emitted in the annealing twin, probably asthe result of strain accommodation (Fig. 9c). As a matter offact, no dislocation can be observed in the adjacent grain, but

the disturbed diffraction contour ahead of the twin tip is anevidence of stress concentration in G2 (Fig. 9d). It can besuspected that upon further straining, stress concentrationin G2 will lead to deformation transmission by emission ofdislocations. The evaluation of the back stress that theannealing twin exerts on the twinning dislocation pile-upcan be evaluated in first approximation by [49]: τ = (nμb / πlK),where μ = 76 GPa is the shear modulus, n = 81 the numberof dislocations, b = 0.144 nm the length of a twinning dislo-cation Burgers vector, l = 1.3 μm, the pile-up length and K =1 − v = 0.69 for edge and K = 1 for screw dislocations.Although the Burgers vector of twinning dislocation has notbeen determined, the dislocation is thought to be b = 1

6 211� �

corresponding to the dislocation with the highest Schmidfactor (0.33) in the 111

� �plane [50]. In this condition, the

dislocation has a pronounced screw character which leads toτ ~200 MPa. This value is of the order of magnitude of thevaluemeasured above for dislocation emission. This indicatesthat annealing twins are strong obstacles to dislocationmotion and thus promote strain hardening inside grains.Deformation propagation is then thought to occur by activa-tion of sources from grain to grain, as observed above. The

author's personal copy

Fig. 8 – In-situ TEM investigation of plastic deformation in the grain shown in Fig. 7c close to a crack tip (see text for detail).

134M

AT

ER

IA

LS

CH

AR

AC

TER

IZ

AT

IO

N94

(2014)

126–137

author's personal copy

Fig. 9 – In-situ TEM image picturing twinning deformation mechanism within NC grains.

135M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

observation of dislocation motion in annealing twin isprobably an efficient way to transfer strain and retardcracking. This process is also thought to be associated tovery low thermal activation volume, i.e. high strain ratesensitivity [51]. These two characteristics of annealing twin,high strength and load transfer availability, are in line withobservations of improved ductility and strength in nano-twinned copper [52].

Annealing twin boundaries are however special bound-aries and the observations described above are only partlyrepresentative of deformation mechanisms occurring in theFG component. Most of the GBs are non-equilibrium HAGBswith a higher energy, and are thought to be “softer” comparedto twin boundaries [53]. In such GB, deformation can howeveralso be accommodated by incorporation and “delocalization”of intergranular dislocations, which is also a high strain ratesensitive mechanism [44,53].

4. Conclusions

SPS route was used to process multimodal bulk nickel samplesfrom powder blends. Tensile tests showed mechanical strength

enhancement (470 MPa) compared to conventional CG nickel butyielded moderate (macroscopic) plastic deformation of about2.8%. Debonding at interfaces between thematrix and the coarseentities was observed probably as a consequence of largedeformation incompatibilities due to a large difference in thegrain size. Indeed, no crack and/or debonding were observedbetween the FG matrix and the FG rim as the average grain sizewas of the same order of magnitude. Therefore control of grainsize gradient between the core and the shell appears to be ofgreat importance for tensile ductility optimisation. Focussed onthe FG matrix, in-situ TEM experiments showed that dislocationemission and annihilation at grain boundaries and twinning atcrack tip were the main deformation mechanisms taking placewithin the FG component.

Supplementary data to this article can be found online athttp://dx.doi.org/10.1016/j.matchar.2014.05.008.

Acknowledgements

This work was carried out in the framework of the “MIMIC”project funded by the French National Research Agency,ANR-09-BLAN-0010.

author's personal copy

136 M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

R E F E R E N C E S

[1] Kumar KS, Van Swygenhoven H, Suresh S. Mechanicalbehavior of nanocrystalline metals and alloys. Acta Mater2003;51:5743–74.

[2] Meyers MA, Mishra A, Benson DJ. Mechanical properties ofnanocrystalline materials. Prog Mater Sci 2006;51:427–556.

[3] Valiev RZ, Langdon TG. Principles of equal-channel angularpressing as a processing tool for grain refinement. Prog MaterSci 2006;51:881–981.

[4] Saada G, Dirras G. The elastic–plastic transition innanograined polycrystals. Int J Mater Res 2009;100:767–9.

[5] [Chapter 90]Saada G, Dirras G. Mechanical properties ofnanograined metallic polycrystals. In: Hirth JP, Kubin L,editors. Dislocations in solids. Elsevier; 2009. p. 199–248.

[6] Wang YM, Ma E. Three strategies to achieve uniform tensiledeformation in a nanostructured metal. Acta Mater2004;52:1699–709.

[7] Van Swygenhoven H, Caro A, Farkas D. Amolecular dynamicsstudy of polycrystalline fcc metals at the nanoscale: grainboundary structure and its influence on plastic deformation.Mater Sci Eng A 2001;309–0:440–4.

[8] Schiøtz J. Atomic-scale modeling of plastic deformation ofnanocrystalline copper. Scr Mater 2004;51:837–41.

[9] Jian WW, Cheng GM, Xu WZ, Yuan H, Tsai MH, Wang QD,et al. Ultrastrong Mg alloy via nano-spaced stacking faults.Mater Res Lett 2013;1:61–6.

[10] Lu L, Shen Y, Chen X, Qian L, Lu K. Ultrahigh strengthand high electrical conductivity in copper. Science2004;304:422–6.

[11] Dao M, Lu L, Asaro RJ, De Hosson JTM, Ma E. Toward aquantitative understanding of mechanical behavior ofnanocrystalline metals. Acta Mater 2007;55:4041–65.

[12] Höppel HW, Zhou ZM, Mughrabi H, Valiev RZ. Microstructuralstudy of the parameters governing coarsening and cyclicsoftening in fatigued ultrafine-grained copper. Phil Mag A2002;82:1781–94.

[13] Dirras G, Gubicza J, Ramtani S, Bui QH, Szilágyi T.Microstructure and mechanical characteristics of bulkpolycrystalline Ni consolidated from blends of powders withdifferent particle size. Mater Sci Eng A 2010;527:1206–14.

[14] Dutel GD, Tingaud D, Langlois P, Dirras G. Nickelwith multimodal grain size distribution achieved by SPS:microstructure and mechanical properties. J Mater Sci2012;47:7926–31.

[15] Srinivasarao B, Oh-ishi K, Ohkubo T, Hono K.Bimodally grained high-strength Fe fabricated by mechanicalalloying and spark plasma sintering. Acta Mater2009;57:3277–86.

[16] Langlois P, Fagnon N, Dirras G. Microstructure engineeringfrom metallic powder blends for enhanced mechanicalproperties. J Phys Conf Ser 2010;240:012016.

[17] Lee Z, Radmilovic V, Ahn B, Lavernia E, Nutt S. Tensiledeformation and fracture mechanism of bulk bimodalultrafine-grained Al–Mg alloy. Metall Mater Trans A2010;41:795–801.

[18] Estrin Y, Vinogradov A. Extreme grain refinement by severeplastic deformation: a wealth of challenging science. ActaMater 2013;61:782–817.

[19] Bui QH, Dirras G, Ramtani S, Gubicza J. On the strengtheningbehavior of ultrafine-grained nickel processed fromnanopowders. Mater Sci Eng A 2010;527:3227–35.

[20] Hemker KJ, Last HR. Microsample tensile testing of LIGA Nifor MEMS applications. Mater Sci Eng A 2001;319–321:882–6.

[21] Buchheit TE, LaVan DA, Michael JR, Christenson TR, Leith SD.Microstructural and mechanical properties investigation ofelectrodeposited and annealed LIGA nickel structures. MetallMater Trans A 2002;33A:539–54.

[22] Cho HS, Hemker KJ, Lian K, Goettert J, Dirras G.Measured mechanical properties of LIGA Ni structures.Sensors Actuators A Phys 2003;103:59–63.

[23] Couque H, Boulanger R. Explosively formed projectilesimulations with Johnson–Cook models. In: Gálvez F,Sánchez-Gálvez V, editors. 23rd International Symposium onBallistic, Tarragona, Spain; 2007. p. 255–62.

[24] Tepper F. Electro-explosion of wire produces nanosizemetals. Met Powder Rep 1998;53:31–3.

[25] Wolff C, Mercier S, Couque H, Molinari A. Modeling ofconventional hot compaction and Spark Plasma Sinteringbased on modified micromechanical models of porousmaterials. Mech Mater 2012;49:72–91.

[26] Wolff C. Modélisation du processus thermo-électro-mécaniquede frittage flash. [PhD Thesis] Metz University; 2011.

[27] Veleva L, Schäublin R, Ramar A, Oksiuta Z, Baluc N. FocusedIon Beam application on the investigation of tungsten-basedmaterials for fusion application. In: Richter S, Schwedt A,editors. EMC 2008 14th European Microscopy Congress 1–5September, Aachen, Germany. Berlin Heidelberg: Springer;2008. p. 503–4.

[28] Coujou A, Lours P, Roy NA, Caillard D, Clement N.Determination of the local tensile axis direction in a tem insitu strained γ′ single crystal—a finite element approach.Acta Metall Mater 1990;38:825–37.

[29] Dutel GD. Comportement Mécanique et Mécanismes deDéformation et d'Endommagement de polycristaux deNickel mono- et bi-modaux élaborés par SPS. [PhD Thesis],13University of Paris; 2013.

[30] Orrù R, Licheri R, Locci AM, Cincotti A, Cao G. Consolidation/synthesis of materials by electric current activated/assistedsintering. Mater Sci Eng R Rep 2009;63:127–287.

[31] Billard S, Fondère JP, Bacroix B, Dirras GF. Macroscopic andmicroscopic aspects of the deformation and fracturemechanisms of ultrafine-grained aluminum processed by hotisostatic pressing. Acta Mater 2006;54:411–21.

[32] Zhang Z, Vajpai SK, Orlov D, Ameyama K. Improvement ofmechanical properties in SUS304L steel through the controlof bimodal microstructure characteristics. Mater Sci Eng A2014;598:106–13.

[33] Dirras G, Tingaud D, Csiszár G, Gubicza J, Couque H, MompiouF. Characterization of bulk bimodal polycrystalline nickeldeformed by direct impact loadings. Mater Sci Eng A2014;601:48–57.

[34] Zhu L, Shi S, Lu K, Lu J. A statistical model for predictingthe mechanical properties of nanostructured metals withbimodal grain size distribution. Acta Mater 2012;60:5762–72.

[35] Orlov D, Fujiwara H, Ameyama K. Obtaining copper withharmonic structure for the optimal balance ofstructure–performance relationship. Mater Trans2013;54:1549–53.

[36] Vajpai SJ, Ota M, Watanabe T, Maeda R, Sekiguchi T, KusakaT, et al. The development of high performance Ti–6Al–4Valloy via a unique microstructural design with bimodal grainsize distribution. Metall Mater Trans A 2014 [submitted forpublication].

[37] Fujiwara H, Akada R, Noro A, Yoshita Y, Ameyama K.Enhanced mechanical properties of nano/meso hybridstructure materials produced by hot roll sintering process.Mater Trans 2008;29:90–6.

[38] Sekiguchi T, Ono K, Fujiwara H, Ameyama K. Newmicrostructure design for commercially pure titanium withoutstanding mechanical properties by mechanical millingand hot roll sintering. Mater Trans 2010;51:39–45.

[39] Douin J, Veyssière P, Beauchamp P. Dislocation line stabilityin Ni3Al. Phil Mag A 1986;54:375–93.

[40] Couret A, Crestou J, Farenc S, Molenat G, Clement N, CoujouA, et al. In situ deformation in T.E.M.: recent developments.Microsc Microanal Microstruct 1993;4:153–70.

author's personal copy

137M A T E R I A L S C H A R A C T E R I Z A T I O N 9 4 ( 2 0 1 4 ) 1 2 6 – 1 3 7

[41] Martin JL, Kubin LP. Discussion on the limitations of “in situ”deformation experiments in a high voltage electronmicroscope. Ultramicroscopy 1978;3:215–26.

[42] Legros M. In situ mechanical TEM: seeing and measuringunder stress with electrons. C R Phys 2014;15:224–40.

[43] Wang Z, McCabe RJ, Ghoniem N, LeSar R, Misra A, Mitchell TE.Dislocation motion in thin Cu foils: comparison betweencomputer simulations and experiment. Acta Mater2004;52:1535–42.

[44] Mompiou F, Legros M, Boé A, Coulombier M, Raskin JP,Pardoen T. Inter- and intragranular plasticity mechanisms inultrafine-grained Al thin films: an in situ TEM study. ActaMater 2013;61:205–16.

[45] Wang Y, Chen M, Zhou F, Ma E. High tensile ductility in ananostructured metal. Nature 2002;419:912–5.

[46] Zhu YT, Liao XZ, Wu XL. Deformation twinning innanocrystalline materials. Prog Mater Sci 2012;57:1–62.

[47] Zhang XY, Wu XL, Zhu AW. Growth of deformation twins inroom-temperature rolled nanocrystalline nickel. Appl PhysLett 2009;94:121907.

[48] Hirth JP, Lothe J. Theory of Dislocations. New York: JohnWiley & Sons; 1982.

[49] Lu L, Schwaiger R, Shan ZW, Dao M, Lu K, Suresh S.Nano-sized twins induce high rate sensitivity of flow stress inpure copper. Acta Mater 2005;53:2169–79.

[50] Christian J, Mahajan S. Deformation twinning. Prog Mater Sci1995;39:1–157.

[51] Lu L, Chen X, Huang X, Lu K. Revealing the maximumstrength in nanotwinned copper. Science 2009;323:607–10.

[52] Priester L. Grain boundaries and crystalline plasticity. Wiley,John & Sons; 2011.

[53] Mompiou F, Caillard D, Legros M, Mughrabi H. In situ TEMobservations of reverse dislocation motion upon unloading intensile-deformed UFG aluminium. Acta Mater 2012;60:3402–14.